속성

Tip

장점

단점

Big-O 계산

방식

hierarchical_clustering_method

hierarchical_clustering_method

상향식 군집화 Agglomerative

전체 객체끼리 가까운 애들끼리 묶기 시작

알고리즘

alggromerative_method

alggromerative_method

single linkage

  • 각 군집에 속한 개체중 가장 작은 거리끼리 묶음
  • 원이 아닌 군집도 찾을 수 있음

complete linkage

Group Average distance

  • k-means와 비슷한 결과가 나올 수 있음

between centroids distance

  • k-means와 비슷한 결과가 나올 수 있음

wards’ method

  • 단순히 거리 정보가 아니라 분산을 고려
  • 두 집합을 합칠 때 사용
  • 집합이 합쳐짐으로써, 군집 내 객체 사이의 분산이 증가한 정도를 확인하여, 증가량이 적은 집합끼리 먼저 합칩 wards_method

하향식 군집화 divisive

  • 전체를 하나의 군집으로 보고 시작
  • 하나의 군집을 n개로 나누는 것을 반복

R 실습

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